Unlock instant, AI-driven research and patent intelligence for your innovation.

monocular vision three-dimensional reconstruction method based on an improved Sift algorithm

A 3D reconstruction and monocular vision technology, applied in computing, image data processing, instruments, etc., can solve the problems of insufficient display of scene structure, shooting angle, precision constraints of feature extraction, unsatisfactory 3D reconstruction results, etc. Achieve good 3D reconstruction effect and improve matching accuracy

Inactive Publication Date: 2019-06-11
HARBIN UNIV OF SCI & TECH
View PDF5 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the constraints of the shooting angle and the accuracy of feature extraction, the 3D reconstruction results are unsatisfactory and cannot fully show the structure of the scene.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • monocular vision three-dimensional reconstruction method based on an improved Sift algorithm
  • monocular vision three-dimensional reconstruction method based on an improved Sift algorithm
  • monocular vision three-dimensional reconstruction method based on an improved Sift algorithm

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0030] Example 1: See Figure 1-6 , a monocular vision three-dimensional reconstruction method based on the improved Sift algorithm, the specific process is as follows: step A, camera calibration. Study several calibration methods of the camera, and choose the most suitable calibration calculation method. Select a high-precision calibration plate and take images of the calibration plate from multiple angles. Detect the feature points in the image respectively, and then calculate the camera internal parameters and camera external parameters according to the selected calibration calculation method to complete the camera calibration;

[0031] Step B, image feature extraction and matching. Select appropriate image features, study image feature extraction methods and image feature description methods, obtain image target feature point sets, and then perform feature matching. Carry out software design, use Halcon image processing software, write programs, complete the realization...

Embodiment 2

[0034] Embodiment 2, on the basis of embodiment 1, improve the SIFT algorithm step as follows:

[0035] Step 1: Use a monocular camera to collect images from multiple angles, apply SIFT operator to any two target images and images to detect extreme points, and filter the obtained feature points to obtain the set sum of SIFT feature points;

[0036] Step 2: Calculate the phase consistency energy information matrix;

[0037] Step 3: Set the threshold value to a value between 0.01 and 0.05, compare it with the calculated phase consistency energy information matrix, and eliminate the feature point pairs whose maximum moment is lower than the threshold value, and obtain the set sum of feature point pairs to be matched;

[0038] Step 4: Perform stereo matching. Calculate the ratio of the closest point to the next closest point of the Euclidean distance and compare it with the set threshold to judge the similarity of the SIFT feature point pair. If the ratio is less than the set th...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a monocular vision three-dimensional reconstruction method based on an improved Sift algorithm. The monocular vision three-dimensional reconstruction method comprises the following steps: A, calibrating a camera; B, performing image feature extraction and matching; C, performing feature extraction and SIFT algorithm improvement; D, performing three-dimensional reconstruction; On the basis of the traditional three-dimensional reconstruction technology, the feature extraction and matching algorithm is improved, and the matching precision is improved, so that a better three-dimensional reconstruction effect is obtained.

Description

technical field [0001] The invention relates to a monocular vision three-dimensional reconstruction method, in particular to a monocular vision three-dimensional reconstruction method based on an improved Sift algorithm. Background technique [0002] The objects in the objective world are three-dimensional, and the images we acquire with cameras are two-dimensional, but we can perceive the three-dimensional information of the target through the two-dimensional images. 3D reconstruction technology processes images in a certain way to obtain 3D information that can be recognized by a computer, and then analyzes the target. Obtaining the model of the real scene from the image is an important research content of computer vision. By reconstructing the 3D model of the target, the digital 3D data can be obtained. These have high application value and broad market prospects. [0003] A complete 3D reconstruction system usually includes three parts: feature matching, camera calibrat...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/80G06T7/55
Inventor 曹宇闫爱宽
Owner HARBIN UNIV OF SCI & TECH